308 ' METALLURGY 03? IRON AND STEEL.
of metal of somewhat finer grain "but much less homogeneous. Considerable care must be exercised in heating pieces which are not to be machined after treatment, since at a high temperature the carbon near the surface of steel is burned out to an appreciable depth by the action of the name, unless the metal is protected in some way from oxidation. An effect of this kind may be noticed under the microscope with little difficulty. If the carbon has been driven off it follows that there is less cementite left to combine with f errite to form pearlite when the metal is cooling through the critical point. Consequently there will be less pearlite formed in the oxidized surface than in the remainder of the piece. This effect is shown in Nos. 38 and 39, these being the center and the outside respectively of a soft steel bar.
In No. 11, Fig. XV-C, is shown a large pearlite grain surrounded by a thick wall of f errite. This represents the micro-structure of a 28-inch steel roll casting containing .25 per cent, carbon and 3.5 per cent, nickel, which was put in service unannealed and broke within a few hours. In No. 10 is shown the fracture in natural size, and the photograph was made from the broken specimen without any polishing or other treatment. It is a striking illustration of intergraniilar weakness, the lines of rupture following almost entirely the ferrite envelope and leaving the individual grains intact. No. 12 shows the micro-structure of this broken roll after one annealing at 800°, and notwithstanding the exceedingly coarse structure of the original casting" the annealed micro-structure is quite fine and shows a grain outline very much broken up. It is probable that a second annealing would have almost obliterated the crystallization, and it would have been interesting to carry this on for several more heat treatments^ [but as this was impracticable a piece was cut off and heated successively to 850°, 800° and 750° Centigrade and allowed to cool slowly with a complete destruction of crystallization as shown in No. 13.
It should be noted that No. 11 and No. 12 are results obtained with full size pieces, and not with small tests, as is too often the case, under which circumstances the results are not always comparable with the effect on a large piece. The two pieces were taken from the same relative positions and represent, it is believed, the structure of the roll. The casting conditions, so far as could be determined, were normal. The annealing was effected at 800° C. as registered by the pyrometer, it being necessary to consider that